Quartz 4

Gemini Dynamo

1 min read

Setup
  • frontend
npm create vite@latest dynamocards
#select React > JavaScript
cd dynamocards
npm install
 
#testing vite
npm run dev
Backend

In backend/main.py:

  • @app.post(“/analyze_video”) def analyze_video(request: VideoAnalysisRequest):
  • Call YoutubeProcessor class from genai.py In backend, make genai.py and __init__.py
  • In genai.py
    • YoutubeProcessor class
      • retrieve_youtube_documents function
        • YoutubeLoader.from_youtube_url receives a youtube link (we use pydantic to check the format) and gets its transcripts (subtitles).
        • We use RecursiveCharacterTextSplitter to split the received input into chunks.

      • find_key_concepts function

    • GeminiProcessor class
      • generate_documents_summary function:Summarize documents using LangChain’s load_summarize_chain
      • count_total_tokens function
      • get_model function
Using middleware

In backend/main.py:

  • Use FastAPI’s middleware: CORSMiddleware→ Processes the requests and responses as they pass through the application. Configures your backend to accept requests from different origins, like your frontend
    • frontend (npm run dev) goes to http://localhost:5173/
    • backend (unicorn main:app —reload) goes to http://127.0.0.1:8000 In frontend/App.jsx
  • empty the file
  • install axios
    • axios →The frontend uses axios to send a POST request with this YouTube link to the backend API endpoint (e.g., /analyze_video).
  • Create a frontend
    • App() using state hooks
    • handles user input link
    • returns the metadata for the link from the backend

Graph View